Hrev_master [page 14] [Emergency Care Journal 2014; 10:1860] Biomarker validation in the emergency department. General criteria and clinical implications Giuseppe Lippi,1 Camilla Mattiuzzi,2 Gianfranco Cervellin3 1Laboratory of Clinical Chemistry and Hematology, Pathology and Laboratory Medicine Department, Parma University Hospital; 2Clinical Governance Service, Trento General Hospital; 3Emergency Department, Parma University Hospital, Italy Abstract The routine use of biomarkers, which is also rapidly expanding in the emergency depart- ment, carries some potential drawbacks such as the risk of producing false positive results and also places a substantial economical bur- den on the healthcare system, especially when the use of laboratory resources is poorly discre- tionary or even inappropriate. The aim of this article is to provide an overview about some general criteria for biomarker validation in the emergency department, and discuss some rel- evant clinical implications. The leading aspects include analysis of data distribution and diagnostic performance, along with eco- nomical and organizational issues. We also brought a pragmatic example, comparing crea- tine kinase MB, a contemporary-sensitive tro- ponin I and a high-sensitivity troponin I immunoassays for evaluation of patients with suspected acute myocardial infarction at emer- gency department admission. Introduction A biological marker, also conventionally known as biomarker, is typically defined as a characteristic that can be measured and used as an indicator of biology, of a pathological process, or even as a guide for targeting specif- ic therapeutic interventions.1,2 Several lines of evidence now attest that the routine use of bio- markers for screening, diagnosis, prognostica- tion, therapeutic monitoring and follow-up of the vast majority of human disorders is virtual- ly unalienable. The greater advantages of bio- marker assessment entails the scarce invasiv- ity wherein a collection of venous blood is only required for performing an impressive number of tests, the short turnaround time (most labo- ratory tests can be performed in less than 30 min), the possibility to process large volumes of analyses due to automation of testing, the relatively low cost as compared with other diagnostic investigations (e.g., diagnostic imaging), and the objective interpretation of data since transversal (i.e., against a specific reference range) or longitudinal (i.e., against previous patient data) comparison of test results is an objective means for establishing whether a given parameter is diagnostic or not. The routine use of biomarkers is rapidly expanding in all fields of science and medi- cine, including the emergency department (ED).3-5 Incidentally, biomarker testing finds its natural application in this peculiar health- care setting, because it provides a rapid means for assessing patients in a growingly over- crowded environment, the collection of blood specimens does not require specific skills, the performance and interpretation of test results does not conventionally require the interven- tion or assistance of other physicians as for radiological examinations.6 In the ED, bio- markers can be used for achieving a final diag- nosis of disease and, even most frequently, for ruling out a clinical suspicion and thereby allowing a safe discharge of patients.7 After that said, the increasingly use of bio- markers in the ED carries also some potential drawbacks, such as the risk of producing false positive results due to the statistics used for establishing reference ranges,8 and also places a substantial economical burden on ED and laboratory, especially when the use of laborato- ry resources is poorly discretionary, or even inappropriate. Therefore, the aim of this arti- cle is to provide an overview about some gen- eral criteria for biomarker validation in the ED, and discuss some relevant clinical implica- tions. For a pragmatic interpretation of several concepts, we will use the emblematic example of acute myocardial infarction (AMI) diagnos- tics at patient admission in the emergency room. This challenging paradigm is particular- ly suited for the topic of this article, since it represents the leading cause of ED admission, and the diagnostic work up is now largely dependent upon results of biomarker test- ing.9,10 The data used in this article were obtained in a population of 98 consecutive patients (mean age 67 years, range: 35-84; 67 males and 31 females) admitted to the ED of the Academic Hospital of Parma for suspected AMI over three working days. Blood samples were collected at patient admission in primary blood tubes containing no additives (Becton Dickinson, Franklin Lakes, NJ, USA) and were rapidly transported to the central laboratory, where they were centrifuged at 1300 ¥ g for 10 min at room temperature. The serum was sep- arated and stored in aliquots for further test- ing. For direct validation and comparison of diagnostic biomarkers, we decided to measure i) an old and virtually obsolete test such as cre- atine kinase isoenzyme MB (CK-MB; Beckman Coulter, Brea, CA, USA), which is character- ized by a limit of detection (LOD) of 0.1 μg/L and a reference range comprised between 0.6 and 6.3 μg/L; ii) a contemporary-sensitive tro- ponin I (TnI) test (AccuTnI; Beckman Coulter), which is characterized by a LoD of 10 ng/L and a 99th percentile of the upper refer- ence limit (URL) of 56 ng/L; and iii) a high- sensitivity (HS) TnI test (HS-AccuTnI; Beckman Coulter), which represent the cur- rent gold standard for diagnosing myocardial injury according to most guidelines, and is characterized by a LoD of 2.5 ng/L and a 99th percentile of the URL of 32 ng/L.11 All biomark- ers were measured on the automated platform Access 2 (Beckman Coulter). A final diagnosis of AMI in our study population could be estab- lished in 11/98 patients (i.e., 11%), according to the well established criteria of the third uni- versal definition of myocardial infarction.12 The statistical analysis was performed with Analyse-it for Microsoft Excel (Analyse-it Software Ltd., Leeds, UK) and MedCalc Version 12.3.0 (MedCalc Software, Mariakerke, Belgium). All ED patients provided an informed consent for participating in this study, which was performed in agreement with the ethical standards established by the insti- tution in which the experiments were per- formed and the Helsinki Declaration of 1975. Opinion Report Analysis of values distribution The foremost step that must be undertaken before performing other types of statistical evaluation is to establish the pattern of data distribution (i.e., Gaussian or not), which then Emergency Care Journal 2014; volume 10:1860 Correspondence: Giuseppe Lippi, Laboratory of Clinical Chemistry and Hematology, Pathology and Laboratory Medicine Department, Parma University Hospital, via Gramsci 14, 43126 Parma, Italy. Tel. +39.0521.703050 - Fax: +39.0521.703791. E-mail: glippi@ao.pr.it, ulippi@tin.it Key words: emergency department, biomarkers, validation, troponin, acute myocardial infarction. Received for publication: 1 August 2013. Revision received: 22 August 2013. Accepted for publication: 8 November 2013. This work is licensed under a Creative Commons Attribution 3.0 License (by-nc 3.0). ©Copyright G. Lippi et al., 2014 Licensee PAGEPress, Italy Emergency Care Journal 2014; 10:1860 doi:10.4081/ecj.2014.1860 Non -co mmerc ial markers in the ED carries also some potential Non -co mmerc ial markers in the ED carries also some potential drawbacks, such as the risk of producing false Non -co mmerc ial drawbacks, such as the risk of producing false positive results due to the statistics used for Non -co mmerc ial positive results due to the statistics used for establishing reference ranges, Non -co mmerc ial establishing reference ranges,8 Non -co mmerc ial 8 and also places Non -co mmerc ial and also places a substantial economical burden on ED and Non -co mmerc ial a substantial economical burden on ED and laboratory, especially when the use of laborato- Non -co mmerc ial laboratory, especially when the use of laborato- ry resources is poorly discretionary, or even Non -co mmerc ial ry resources is poorly discretionary, or even inappropriate. Therefore, the aim of this arti- Non -co mmerc ial inappropriate. Therefore, the aim of this arti- Non -co mmerc ial immunoassays for evaluation of patients with Non -co mmerc ial immunoassays for evaluation of patients with suspected acute myocardial infarction at emer- Non -co mmerc ial suspected acute myocardial infarction at emer- Non -co mmerc ial cle is to provide an overview about some gen- Non -co mmerc ial cle is to provide an overview about some gen- eral criteria for biomarker validation in the ED, Non -co mmerc ial eral criteria for biomarker validation in the ED, and discuss some relevant clinical implica- Non -co mmerc ial and discuss some relevant clinical implica- tions. For a pragmatic interpretation of several Non -co mmerc ial tions. For a pragmatic interpretation of several concepts, we will use the emblematic example Non -co mmerc ial concepts, we will use the emblematic example of acute myocardial infarction (AMI) diagnos- Non -co mmerc ial of acute myocardial infarction (AMI) diagnos- us e ruling out a clinical suspicion and thereby us e ruling out a clinical suspicion and thereby After that said, the increasingly use of bio-us e After that said, the increasingly use of bio- markers in the ED carries also some potentialus e markers in the ED carries also some potential Coulter, Brea, CA, USA), which is character- us e Coulter, Brea, CA, USA), which is character- ized by a limit of detection (LOD) of 0.1 us e ized by a limit of detection (LOD) of 0.1 and a reference range comprised between 0.6 us e and a reference range comprised between 0.6 on ly atine kinase isoenzyme MB (CK-MB; Beckmanon ly atine kinase isoenzyme MB (CK-MB; Beckman Coulter, Brea, CA, USA), which is character-on ly Coulter, Brea, CA, USA), which is character-on ly Emergency Care Journal 2014; 10:1860 on ly Emergency Care Journal 2014; 10:1860 doi:10.4081/ecj.2014.1860 on lydoi:10.4081/ecj.2014.1860 [Emergency Care Journal 2014; 10:1860] [page 15] influences the selection of the most appropri- ate statistical approach. This can be accom- plished with several tests, the most widely used are represented by Kolmogorov-Smirnov, D’Agostino-Pearson, Anderson-Darling and Shapiro-Wilk W tests. When the results of these tests (i.e., the P value) are higher than 0.05, it can be assumed that data follow a nor- mal (Gaussian) distribution, whereas in the presence of P values lower than 0.05 the hypothesis that the distribution of data is nor- mal should be rejected.13 The analysis of our patient population, using the Kolmogorov-Smirnov test for nor- mality, is shown in Figure 1, which clearly demonstrates that the values distribution of none of the three biomarkers follows a normal pattern, so that a non parametric approach must be used for their further analysis. This typically entails that patient values should be displayed in terms of median and percentile (or interquartile range; IQR) rather than in terms of mean and standard deviation, unless values are logarithmically transformed to achieve a normal distribution. Accordingly, the Wilcoxon-Mann-Whitney test (rather than the Student’s t test, which is used for data normal- ly distributed) is the preferred approach for distinguishing whether the results obtained in the populations with or without disease (i.e., AMI) are significantly different. The results of biomarkers measurement in our study popula- tion are reported in Table 1, which shows that a significant difference exists between patients with or without a final diagnosis of AMI. This analysis is a necessary premise for establishing whether a biomarker may be clin- ically usable or not, because the lack of signif- icant differences of values between patients with disease and those without would also imply a lack of real clinical usefulness. A graphical representation of data may also be useful under some circumstances, as shown in Figure 2, but this is not strictly necessary. Receiver operating characteristic curve analysis Once the analysis of raw data has been com- pleted as described in the former paragraph, the second advisable step entails the evalua- tion of diagnostic performance of the test(s). The preferred means is indeed represented by the use of receiver operating characteristic (ROC) curve analysis. This approach allows to obtain a graphical representation of the overall diagnostic performance, which is typically quantified in terms of Area Under the Curve, AUC. The closest is the value of the AUC to 1.0, the better is the diagnostic performance. The output of the analysis is also supported by a degree of statistical significance (i.e., P value) and a complete description of diagnostic sensi- tivity, specificity, negative predictive value (NPV), positive predictive value (PPV), nega- tive likelihood ratio (LR-) and positive likeli- hood ratio (LR+). The local calculation of these parameters, when feasible, is indeed preferable over the use of manufacturer’s ref- erence rages or predefined diagnostic thresh- olds established on the 99th percentile of the URL, since these limits have been calculated on populations that may largely differ from those typically referred to the local ED.14 Indeed, ROC curves are very useful, but not completely pervasive in trials including a mod- est sample size. Nevertheless, this type of analysis is indeed the core for assessing the diagnostic performance of biomarkers and rep- resents a suitable background for planning larger and more focused investigations, by Opinion Report Table 1. Results of biomarker measurement in a population of 98 consecutive patients admitted to the emergency department with suspected acute myocardial infarction. Non-AMI (n=87) AMI (n=11) P CK-MB (μg/L) 1.9 (1.2-2.8) 7.5 (5.5-16.7) <0.001 AccuTnI (ng/L) 3 (0-20) 900 (35-1373) <0.001 HS-AccuTnI (ng/L) 7 (4-23) 621 (41-1237) <0.001 AMI, acute myocardial infarction; CK-MB, creatine kinase isoenzyme MB; AccuTnI, contemporary-sensitive Troponin I immunoassay; HS- AccuTnI, high-sensitivity Troponin I immunoassay. Results are shown as median and interquartile range and the significance of differences is assessed with Wilcoxon-Mann-Whitney test. Figure 1. Value distribution of creatine kinase isoenzyme MB, a contemporary-sensitive Troponin I immunoassay and a high-sensitivity Troponin I immunoassay in a population of 98 consecutive patients admitted to the emergency department with suspected acute myocar- dial infarction. The assumption of normality is tested with Kolmogorov-Smirnov test. Non -co mmerc ial icant differences of values between patients Non -co mmerc ial icant differences of values between patients with disease and those without would also Non -co mmerc ial with disease and those without would also imply a lack of real clinical usefulness. A Non -co mmerc ial imply a lack of real clinical usefulness. A graphical representation of data may also be Non -co mmerc ial graphical representation of data may also be useful under some circumstances, as shown in Non -co mmerc ial useful under some circumstances, as shown in Figure 2, but this is not strictly necessary. Non -co mmerc ial Figure 2, but this is not strictly necessary. Receiver operating characteristicNon -co mmerc ial Receiver operating characteristicNon -co mmerc ial us e o nly [page 16] [Emergency Care Journal 2014; 10:1860] determining decision thresholds and prelimi- nary clinical usefulness.15 According to the diagnostic performance, ROC curves thus allow to identify the optimal cut-off, that is the biomarker value associated with the highest result of the sum of sensibility and specificity. However, this value does not necessary repre- sent the best threshold according to the intend- ed clinical use of the test. The assessment of D-dimer for ruling out venous thromboem- bolism (VTE) is a typical example. According to its validated use, D-dimer testing typically fol- lows the evaluation of pre-test probability of VTE, and its value influences the decision to performed additional imaging investigations (e.g., leg ultrasonography or lung computed tomography).16 As such, the best cut-off in this setting is not the compromise between sensi- tivity and specificity, but rather the D-dimer value characterized by the highest sensitivity (advisably 1.0), since this would allow to avoid additional (invasive) testing and discharge patients. Similar considerations can be drown for the use of proteins S100B for the diagnosis of brain injury in patients with mild head trau- ma, wherein a non diagnostic value of this bio- marker should permit to prevent unnecessary head CT.17 The comparison of the ROC curves of the three biomarkers in our study population is shown in Figure 3. The AUC of the different bio- markers was 0.90 (95% CI, 0.77 to 1.00; P<0.001) for CK-MB, 0.92 (95% CI, 0.85 to 0.99; P<0.001) for AccuTnI and 0.92 (95% CI, 0.84 to 0.99; P<0.001) for HS-AccuTnI. Interestingly, although the AUCs of both TnI immunoassays were better than that of CK-MB, no significant differences were found (i.e., CK-MB vs AccuTnI, P=0.68; CK-MB vs HS-AccuTnI, P=0.69; AccuTnI vs HS-AccuTnI, P=0.77). The optimal thresh- olds, which incidentally corresponds to the best cut-offs for both TnI immunoassays (i.e., those associated with 1.00 sensitivity for rule out of AMI upon ED admission), were 4.6 μg/L for CK- MB, 17 ng/L for AccuTnI, and 14 ng/L for HS- AccuTnI, respectively (Table 2). Although the diagnostic performance of CK-MB was relative- ly poor as compared with those of both TnI immunoassays (i.e., the specificity was higher but the sensitivity was indeed unacceptable), it is noteworthy that, in analogy with recently pub- lished data,18,19 the diagnostic performance of the contemporary-sensitive (i.e., AccuTnI) and HS (i.e., HS-AccuTnI) tests was nearly identical, with a negligible better specificity and PPV of the former method. An additionally useful test is then represented by the calculation of the diag- nostic odds ratio (DOR), which is synthetically defined as the ratio of the odds of the test being positive if the subject has a disease relative to the odds of the test being positive if the subject does not have the disease [i.e., (true positive/false negative)/(false positive/ true negative)]. Also in this circumstance the DORs Opinion Report Figure 2. Value distribution of creatine kinase isoenzyme MB, a contemporary-sensitive Troponin I immunoassay and a high-sensitivity Troponin I immunoassay in a population of 98 consecutive patients admitted to the emergency department with suspected acute myocardial infarction. Results are shown as median and interquartile range and the sig- nificance of differences is assessed with Wilcoxon-Mann-Whitney test. Non -co mmerc ial HS-AccuTnI, P=0.77). The optimal thresh- Non -co mmerc ial HS-AccuTnI, P=0.77). The optimal thresh- olds, which incidentally corresponds to the best Non -co mmerc ial olds, which incidentally corresponds to the best ., those Non -co mmerc ial ., those associated with 1.00 sensitivity for rule out of Non -co mmerc ial associated with 1.00 sensitivity for rule out of AMI upon ED admission), were 4.6 Non -co mmerc ial AMI upon ED admission), were 4.6 μ Non -co mmerc ial μg/L for CK- Non -co mmerc ial g/L for CK- MB, 17 ng/L for AccuTnI, and 14 ng/L for HS-Non -co mmerc ial MB, 17 ng/L for AccuTnI, and 14 ng/L for HS- AccuTnI, respectively (Table 2). Although theNon -co mmerc ial AccuTnI, respectively (Table 2). Although theNon -co mmerc ial us e o nly [Emergency Care Journal 2014; 10:1860] [page 17] of both TnI immunoassays calculated in our study population outperform that of CK-MB, with the DOR of the contemporary assay being slightly higher than that of the HS due to a lower number of false positive cases (Table 2). Organizational and cost-effective- ness analysis A final important evaluation, especially in a world of limited resources, is the organization- al and economical impact that the assessment of the various biomarker poses on the health- care system and the potential savings due to early or more accurate diagnosis.20,21 As regards practical considerations, the turnaround time is the leading aspect to be considered. According to recent guidelines, which entails serial sampling of cardiac bio- markers at 0, 3 and 6 h,22 it is absolutely neces- sary to use analytical techniques that can pro- duce results in less than 60 min, which should also be located at a convenient distance from the ED to avoid delays due to sample trans- portation.23 When both requisites cannot be fulfilled, introduction of point of care testing in the ED may be a viable option.24 As regards the economical burden, there is no simple means for calculating reliable fig- ures, but one suitable approach can be sug- gested. The number needed to test (NNT) can be synthetically described as the number of patients that should be tested to identify one additional adverse outcome (e.g., AMI). Although this calculation is more typically used for defining the number needed to screen (NNS) within policies of risk reduction (e.g., prostate specific antigen screening for pre- venting mortality from prostate cancer), it can also be reliably applied for gathering informa- tion about the organizational and clinical bur- dens that one given test poses on healthcare resources.25 It is rather obvious that the high- er is the value of the NNT, the larger is the cost for diagnosing a given disorder. According to our data, the NTT is expectedly lower for both TnI immunoassays (i.e., 9) as compared with CK-MB (i.e., 11). By translation of this concept into economical terms (i.e., the cost of a single determination is € 0.92 for CK-MB and € 1.12 for AccuTnI in our institution), the final cost of routinely using CK-MB or AccuTnI would result to be nearly identical (i.e., € 10.12 versus € 10.08) (Table 3). Nevertheless, the cost of CK- MB will be then inevitably inflated by the larg- er expenditure that emerges from the consid- erable number of missed diagnoses upon patient admission (2/11; i.e., 18% versus 0/11 with both AccuTnI tests). This cost, which can- not be reliably estimated, is mostly attributable to delayed therapeutic intervention, worse out- comes, longer stay and greater consumption of ED resources. Opinion Report Table 2. Diagnostic performance of creatine kinase isoenzyme MB, a contemporary-sen- sitive Troponin I immunoassay and a high-sensitivity Troponin I immunoassay for diag- nosing acute myocardial infarction in a population of 98 consecutive patients admitted to the emergency department. CK-MB AccuTnI HS-AccuTnI AUC 0.90* 0.92° 0.92# Cut-off 4.6 μg/L 17 ng/L 14 ng/L Sensitivity 0.82 1.00 1.00 Specificity 0.93 0.75 0.72 NPV 0.98 1.00 1.00 PPV 0.60 0.33 0.31 LR- 0.20 0.00 0.00 LR+ 11.86 3.95 3.63 DOR 51§ 60^ 56$ CK-MB, creatine kinase isoenzyme MB; AccuTnI, contemporary-sensitive Troponin I immunoassay; HS-AccuTnI, high-sensitivity Troponin I immunoassay; AUC, area under the curve; NPV, negative predictive value; PPV, positive predictive value; LR-, negative likelihood ratio; LR+, pos- itive likelihood ratio; DOR, diagnostic odds ratio. *95% CI, 0.77 to 1.00; P<0.001; °95% CI, 0.85 to 0.99; P<0.001; #95% CI, 0.84 to 0.99; P<0.001; §95% CI, 9 to 286; P<0.01; ^95% CI, 3 to 1050; P<0.01; $95% CI, 3 to 993; P<0.01. Table 3. Economical analysis of assessing creatine kinase isoenzyme MB, a contemporary- sensitive Troponin I immunoassay and a high-sensitivity Troponin I immunoassay for diagnosing acute myocardial infarction in a population of 98 consecutive patients admit- ted to the emergency department. CK-MB AccuTnI HS-AccuTnI Number needed to test 11 9 9 Cost per test (€) 0.92 1.12 - Total cost (€) 10.12 10.08 - CK-MB, creatine kinase isoenzyme MB; AccuTnI, contemporary-sensitive Troponin I immunoassay; HS-AccuTnI, high-sensitivity Troponin I immunoassay. Figure 3. Receiver operating characteristics curve of creatine kinase isoenzyme MB, a con- temporary-sensitive Troponin I immunoassay and a high-sensitivity Troponin I immunoassay for diagnosing acute myocardial infarction in a population of 98 consecu- tive patients admitted to the emergency department. Non -co mmerc ial Table 2. Diagnostic performance of creatine kinase isoenzyme MB, a contemporary-sen- Non -co mmerc ial Table 2. Diagnostic performance of creatine kinase isoenzyme MB, a contemporary-sen- sitive Troponin I immunoassay and a high-sensitivity Troponin I immunoassay for diag- Non -co mmerc ial sitive Troponin I immunoassay and a high-sensitivity Troponin I immunoassay for diag- nosing acute myocardial infarction in a population of 98 consecutive patients admitted Non -co mmerc ial nosing acute myocardial infarction in a population of 98 consecutive patients admitted to the emergency department. Non -co mmerc ial to the emergency department. Non -co mmerc ial Non -co mmerc ial Non -co mmerc ial Cut-off Non -co mmerc ial Cut-off Non -co mmerc ial also be reliably applied for gathering informa- Non -co mmerc ial also be reliably applied for gathering informa- tion about the organizational and clinical bur- Non -co mmerc ial tion about the organizational and clinical bur- dens that one given test poses on healthcare Non -co mmerc ial dens that one given test poses on healthcare It is rather obvious that the high- Non -co mmerc ial It is rather obvious that the high- er is the value of the NNT, the larger is the cost Non -co mmerc ial er is the value of the NNT, the larger is the cost for diagnosing a given disorder. According toNon -co mmerc ial for diagnosing a given disorder. According to our data, the NTT is expectedly lower for bothNon -co mmerc ial our data, the NTT is expectedly lower for both ., 9) as compared withNon -co mmerc ial ., 9) as compared with Sensitivity Non -co mmerc ial Sensitivity Non -co mmerc ial Non -co mmerc ial Specificity Non -co mmerc ial Specificity NPV Non -co mmerc ial NPV Non -co mmerc ial Non -co mmerc ial PPV Non -co mmerc ial PPV us e Table 2. Diagnostic performance of creatine kinase isoenzyme MB, a contemporary-sen-us e Table 2. Diagnostic performance of creatine kinase isoenzyme MB, a contemporary-sen- sitive Troponin I immunoassay and a high-sensitivity Troponin I immunoassay for diag-us e sitive Troponin I immunoassay and a high-sensitivity Troponin I immunoassay for diag- on ly Figure 3. Receiver operating characteristics curve of creatine kinase isoenzyme MB, a con- on ly Figure 3. Receiver operating characteristics curve of creatine kinase isoenzyme MB, a con- temporary-sensitive Troponin I immunoassay and a high-sensitivity Troponin I on lytemporary-sensitive Troponin I immunoassay and a high-sensitivity Troponin I immunoassay for diagnosing acute myocardial infarction in a population of 98 consecu- on lyimmunoassay for diagnosing acute myocardial infarction in a population of 98 consecu- on lytive patients admitted to the emergency department. on lytive patients admitted to the emergency department. [page 18] [Emergency Care Journal 2014; 10:1860] Conclusions Biomarkers validation is a challenging but necessary enterprise in almost every area of medical sciences, thus including the ED. The indiscriminate introduction of biomarkers in the ED, along with their inappropriate request, carries a high risk of consuming valuable human or economical resources and jeopardiz- ing patient safety. In this article we have dis- cussed some relevant issues that should guide the decision as to whether the introduction (or replacement) of a given biomarker is clinically justified and economically acceptable. According to data obtained using three differ- ent biomarkers (i.e., old, current and innova- tive) for diagnosing AMI in the ED, we could confirm that the use of CK-MB appears now largely unjustified. It is noteworthy, however, that we could not find a real clinical improve- ment using a HS-TnI immunoassay as com- pared with the previous contemporary-sensi- tive method, at least upon patient admission to the ED. Although recent data attests that the former test would outstrip the previous tech- niques during serial testing according to its improved analytical sensitivity, this paradigm has recently been challenged by additional data showing that the diagnostic accuracy of some contemporary sensitive and HS immunoassays may be virtually identical using the recommended 2-3 h sampling protocol.18,26 Despite the fact that we could not provide definitive economical data about the compari- son of AccuTnI versus HS-AccuTnI because the latter test has not become commercially avail- able so far, it is plausible to assume that the expenditure per test of the HS method would be not less than 20% higher (also also con- firmed by the manufacturer). As such, consid- ering that the clinical performance are virtual- ly identical, it should not be ignored that the replacement of a well-suited contemporary- sensitive immunoassay with a novel HS method would be associated with a larger expenditure for the healthcare system. A final mention deserves the interface between the emergency physician, the laboratory profes- sional and the medical direction. These three parties, which are actively involved in the proj- ect of introducing a novel biomarker, should actively cooperate for defining clinical paths, identifying the appropriate settings of imple- mentation, as well as for monitoring organiza- tional, clinical and economical outcomes. Active and forthright collaboration is the key to foster successful relationships and improve efficacy and efficiency in the ED.27 References 1. Lippi G, Plebani M. Biomarker research and leading causes of death worldwide: a rather feeble relationship. Clin Chem Lab Med 2013:51;1691-3. 2. Braunwald E. Biomarkers in heart failure. N Engl J Med. 2008;358:2148-59. 3. Di Somma S, Magrini L, Travaglino F, et al. Opinion paper on innovative approach of biomarkers for infectious diseases and sepsis management in the emergency department. Clin Chem Lab Med 2013;51: 1167-75. 4. Lippi G, Plebani M, Di Somma S, et al. Considerations for early acute myocardial infarction rule-out for emergency depart- ment chest pain patients: the case of copeptin. Clin Chem Lab Med 2012;50:243- 53. 5. Lippi G, Valentino M, Cervellin G. Laboratory diagnosis of acute pancreatitis: in search of the Holy Grail. Crit Rev Cl Lab Sci 2012;49:18-31. 6. Cavazza M. Biomarkers in emergency medicine: great opportunities or expen- sive puzzles? Emerg Care J 2012;8:3-4. 7. Lippi G, Guidi GC. The power of negative thinking. Am J Emerg Med 2008;26:373-4. 8. Lippi G, Plebani M. False myths and leg- ends in laboratory diagnostics. Clin Chem Lab Med 2013;51:2087-97. 9. Lippi G, Franchini M, Cervellin G. Diagnosis and management of ischemic heart disease. Semin Thromb Hemost 2013;39:202-13. 10. Lippi G, Cavazza M, Peracino A, et al. Ischemic heart disease in the emergency room: state of the art, innovation and research. Emerg Care J 2013;9:e7. 11. Apple FS, Ler R, Murakami MM. Determination of 19 cardiac troponin I and T assay 99th percentile values from a com- mon presumably healthy population. Clin Chem 2012;58:1574-81. 12. Thygesen K, Alpert JS, Jaffe AS, et al. Third universal definition of myocardial infarc- tion. Circulation 2012;126:2020-35. 13. Griner PF, Mayewski RJ, Mushlin AI, Greenland P. Selection and interpretation of diagnostic tests and procedures. Ann Intern Med 1981;94:555-600. 14. Lippi G, Margapoti R, Aloe R, Cervellin G. Highly-sensitive troponin I in patients admitted to the emergency room with acute infections. Eur J Intern Med 2013; 24:e57-8. 15. Kampfrath T, Levinson SS. Brief critical review: statistical assessment of biomark- er performance. Clin Chim Acta 2013;419: 102-7. 16. Lippi G, Franchini M, Targher G, Favaloro EJ. Help me, Doctor! My D-dimer is raised. Ann Med 2008;40:594-605. 17. Cervellin G, Benatti M, Carbucicchio A, et al. Serum levels of protein S100B predict intracranial lesions in mild head injury. Clin Biochem 2012;45:408-11. 18. Lippi G, Cervellin G. Do we really need high-sensitivity troponin immunoassays in the emergency department? Maybe not. Clin Chem Lab Med 2013;52:205-11. 19. Lippi G, Cervellin G. Highly-sensitive immunoassays in the emergency depart- ment: counterpoint. Emerg Care J 2013;9: e16. 20. Lippi G, Cervellin G. Letter to the Editor: choosing troponin immunoassays in a world of limited resources. J Am Coll Cardiol 2013;62:647-8. 21. Iannone P. The need of a health technolo- gy assessment perspective in emergency medicine. Emerg Care J 2013;9:e8. 22. Casagranda I, Cavazza M, Clerico A, et al. Proposal for the use in emergency depart- ments of cardiac troponins measured with the latest generation methods in patients with suspected acute coronary syndrome without persistent ST-segment elevation. Clin Chem Lab Med 2013;51:1727-37. 23. Lippi G, Simundic AM, Plebani M. Phlebotomy, stat testing and laboratory organization: an intriguing relationship. Clin Chem Lab Med 2012;50:2065-8. 24. Lippi G, Mattiuzzi C, Cervellin G. Point of care troponin testing: rules and regula- tions. J Electrocardiol 2013;46:727-8. 25. Rembold CM. Number needed to screen: development of a statistic for disease screening. Brit Med J 1998;317:307-12. 26. Lippi G, Cervellin G. Challenges of serial troponin testing: an unfinished symphony. Int J Cardiol 2013;168:4397. 27. Cavazza M. ECJ: a new journal for new challenges in emergency medicine. Emerg Care J 2013;9:e1. Opinion Report Non -co mmerc ial Laboratory diagnosis of acute pancreatitis: Non -co mmerc ial Laboratory diagnosis of acute pancreatitis: in search of the Holy Grail. Crit Rev Cl Lab Non -co mmerc ial in search of the Holy Grail. Crit Rev Cl Lab 6. Cavazza M. Biomarkers in emergency Non -co mmerc ial 6. Cavazza M. Biomarkers in emergency medicine: great opportunities or expen- Non -co mmerc ial medicine: great opportunities or expen- sive puzzles? Emerg Care J 2012;8:3-4. Non -co mmerc ial sive puzzles? Emerg Care J 2012;8:3-4. 7. Lippi G, Guidi GC. The power of negative Non -co mmerc ial 7. Lippi G, Guidi GC. The power of negative thinking. Am J Emerg Med 2008;26:373-4. Non -co mmerc ial thinking. Am J Emerg Med 2008;26:373-4. Non -co mmerc ial be not less than 20% higher (also also con- Non -co mmerc ial be not less than 20% higher (also also con- firmed by the manufacturer). As such, consid- Non -co mmerc ial firmed by the manufacturer). As such, consid- ering that the clinical performance are virtual- Non -co mmerc ial ering that the clinical performance are virtual- ly identical, it should not be ignored that the Non -co mmerc ial ly identical, it should not be ignored that the replacement of a well-suited contemporary-Non -co mmerc ial replacement of a well-suited contemporary- sensitive immunoassay with a novel HSNon -co mmerc ial sensitive immunoassay with a novel HS method would be associated with a largerNon -co mmerc ial method would be associated with a larger 8. Lippi G, Plebani M. False myths and leg- Non -co mmerc ial 8. Lippi G, Plebani M. False myths and leg- ends in laboratory diagnostics. Clin Chem Non -co mmerc ial ends in laboratory diagnostics. Clin Chem Lab Med 2013;51:2087-97. Non -co mmerc ial Lab Med 2013;51:2087-97. 9. Lippi G, Franchini M, Cervellin G. Non -co mmerc ial 9. Lippi G, Franchini M, Cervellin G. Diagnosis and management of ischemic Non -co mmerc ial Diagnosis and management of ischemic heart disease. Semin Thromb Hemost Non -co mmerc ial heart disease. Semin Thromb Hemost us e copeptin. Clin Chem Lab Med 2012;50:243- us e copeptin. Clin Chem Lab Med 2012;50:243- 5. Lippi G, Valentino M, Cervellin G. us e 5. Lippi G, Valentino M, Cervellin G. Laboratory diagnosis of acute pancreatitis:us e Laboratory diagnosis of acute pancreatitis: in search of the Holy Grail. Crit Rev Cl Labus e in search of the Holy Grail. Crit Rev Cl Lab 20. Lippi G, Cervellin G. Letter to the Editor: us e 20. Lippi G, Cervellin G. Letter to the Editor: on ly high-sensitivity troponin immunoassays on ly high-sensitivity troponin immunoassays in the emergency department? Maybe not. on lyin the emergency department? Maybe not. on lyClin Chem Lab Med 2013;52:205-11. on lyClin Chem Lab Med 2013;52:205-11. 19. Lippi G, Cervellin G. Highly-sensitiveon ly19. Lippi G, Cervellin G. Highly-sensitive immunoassays in the emergency depart-on ly immunoassays in the emergency depart- ment: counterpoint. Emerg Care J 2013;9:on ly ment: counterpoint. Emerg Care J 2013;9: